A Pilot Study Evaluating a Simple Cardiac Dysfunction Score to Predict Complications and Survival Among Critically-ill Patients with Traumatic Brain Injury

Abstract

Purpose

To describe the frequency of cardiovascular complications and cardiac dysfunction in critically-ill patients with moderate-severe traumatic brain injury (msTBI) and cardiac factors associated with in-hospital survival.

Methods

Retrospective analysis of a prospective cohort study at a single Level-1 trauma center with a dedicated neuro-trauma intensive care unit (ICU). Adult patients admitted to the ICU with msTBI were consecutively enrolled in the prospective OPTIMISM study between November 2009 and January 2017. Cardiac dysfunction was measured using a combination of EKG parameters, echocardiography abnormalities, and peak serum troponin-I levels during the index hospitalization. These items were combined into a cardiac dysfunction index (CDI), ranging from 0 to 3 points and modeled in a Cox regression analysis.

Results

A total of 326 patients with msTBI were included. For every one-point increase in the CDI, the multivariable adjusted risk of dying during the patient’s acute hospitalization more than doubled (adjusted HR 2.41; 95% CI 1.29–4.53).

Conclusion

Cardiac dysfunction was common in patients with msTBI and independently associated with more severe brain injury and a reduction in hospital survival in this population. Further research is needed to validate the CDI and create more precise scoring tools.

INTRODUCTION

Moderate-severe TBI (msTBI) continues to be the leading cause of death and disability after trauma, accounting for the majority of the 52,000 TBI-related deaths in the U.S. annually [1]. Patients with msTBI are commonly admitted to a specialized neurological or neuro-trauma Intensive Care Unit (neuroICU) where they receive complex, coordinated medical and surgical care. While in the neuroICU, patients are at increased risk for a number of clinically significant medical and neurological complications, which are associated with short- and long-term outcomes after msTBI [2]. For example, cardiovascular dysfunction has been reported in a number of studies examining certain neurological insults and a “Cardio-Cerebral Syndrome” has been described in several retrospective and cross-sectional studies in neuro-critically ill patients.

These studies found cardiac dysfunction, as measured by either echocardiographic abnormalities or elevations in serum troponin findings, to be related to the presence of certain intracranial pathologies, particularly ischemic stroke and subarachnoid hemorrhage [3–8]. The published literature describing brain-heart interactions in patients with TBI is, however, even more limited. A small number of studies have found an increased prevalence of cardiac dysfunction (abnormal EKG or reduced systolic function measured by echocardiography) in the setting of TBI [9–11]. Other studies have attempted to relate cardiac dysfunction to in-hospital mortality [12]. One large retrospective review found an increased myocardial workload and rate-pressure product following severe TBI with increased in-hospital mortality for those with depressed or increased rate-pressure products [13].

What remains largely unknown is how often various cardiovascular complications occur in msTBI patients hospitalized in the neuroICU and whether cardiac biomarker, electrocardiography, and imaging studies can be utilized to predict patient outcomes. In our exploratory study, we examined the frequency of cardiovascular complications among patients hospitalized in a neuroICU with msTBI. Second, we evaluated the relationship between the severity of brain injury and extent of cardiac dysfunction to determine if there exists a type of dose-response relationship in this cardio-cerebral syndrome. Lastly, we assessed the impact of these clinical complications and cardiac parameters on patients’ short-term survival using a simple cardiac dysfunction index (CDI). We hypothesized that increasing severity of TBI would be independently associated with higher degrees of cardiac injury as measured by the CDI and patients who experienced cardiac dysfunction in the setting of their TBI would be less likely to survive their acute hospitalization.

METHODS

Study Population

We conducted a retrospective analysis of prospectively collected data from the Outcome Prognostication in Traumatic Brain Injury (OPTIMISM) study in which patients with msTBI (Glasgow Coma Scale (GCS)≤12) admitted to the neuroICU at the University of Massachusetts Medical School/UMass Memorial Medical Center level-1 trauma center, were consecutively enrolled between November, 2009 and January, 2017. Details of the OPTIMISM methodology have previously been published [2,14]. This study was approved by the Institutional Review Board (IRB) at the University of Massachusetts Medical School.

Clinical Management

As described previously [2], after initial resuscitation in the emergency department (ED), patients were admitted to the neuroICU, staffed by board-certified neuro-intensivists or a trauma-intensivist, and managed according to Brain Trauma Foundation guidelines [15, 16]. All patients with a post-resuscitation GCS ≤8 and evidence of brain swelling or high-risk lesions received an intracranial pressure (ICP) monitoring device (intraparenchymal probe or extraventricular drain) as well as emergency neurosurgery based on clinical indications and at the discretion of the consulting neurosurgeon. After 2012, additional brain tissue oxygen (PbtO₂) monitoring with Licox^© (Integra, Plainsboro, NJ) was performed. Both ICP and PbtO₂ were treated according to an institutional protocol that mimicked the Brain Tissue Oxygen Monitoring in Traumatic Brain Injury (BOOST-2) protocol [17]. Our standard institutional practice is to avoid early withdrawal of life-sustaining treatment in the first 5–7 days of the patient’s hospitalization, unless the patient has irreversible brainstem herniation.

Study Data

Baseline demographic characteristics, trauma mechanism, pupillary reactivity, and presence of hypoxia (oxygen saturation levels ≤89%) or hypotension (systolic blood pressure ≤89 mmHg) in the field or ED were collected. Out-of-hospital and ED data were obtained from written ED trauma attending and nursing reports. The study enrollment GCS was defined as the lowest post-resuscitation GCS during the first 24 hours of hospital admission after sedation or intoxication subsided. Trained trauma registrars calculated the Injury Severity Score (ISS) and patients were followed for the occurrence of 28 predefined medical and neurological complications throughout the ICU stay. These complications were adjudicated on a weekly basis by three neurointensivists [2]. Cardiac ICU complications routinely collected in OPTIMISM include occurrence of a new, not previously existing, cardiac arrhythmia (atrial or ventricular fibrillation, ventricular tachycardia, symptomatic bradycardia), hypotension requiring vasopressor infusion, deep venous thrombosis (DVT), pulmonary embolism (PE), myocardial infarction (MI; defined as new ischemic ECG changes, peak serum troponin-I [sTnI] levels >0.5 ng/mL and confirmation of wall motion abnormalities by echocardiography), and cardiac arrest in the ICU. Information about pre-existing cardiovascular disease (CVD), which included whether patients had been previously diagnosed with coronary artery disease, MI, hypertension, hypercholesterolemia, arrhythmias, heart failure, or hyperlipidemia was also collected.

For the present study, we recorded several additional measurable cardiac parameters by chart review: EKG computer-generated estimates of the ventricular rate, PR interval, QTc interval, R-wave axis, T-wave axis, and QRS interval on the patient’s admission EKG, peak sTnI levels during their index hospitalization, and transthoracic echocardiography (TTE) measurements by a board-certified cardiologist including ejection fraction (%), fractional shortening, and qualitative assessments (e.g., diastolic function and wall motion abnormalities). These measures of cardiac dysfunction were ordered by the treating team when clinically indicated.

We created a clinically meaningful and easy-to-derive CDI which consisted of three equally-weighted components: 1) elevated sTnI (>0.04 ng/mL); 2) development of an EKG abnormality (either prolonged corrected QTc interval, widened QRS, presence of T-wave inversions, or irregularity of the R-R interval); and 3) any TTE abnormalities (reduced ejection fraction (EF <50%), wall motion abnormalities (e.g., global hypokinesis), or fractional shortening). These items were chosen because they can be easily and routinely measured in critically-ill patients and yield information about cardiac function, such as the presence of structural damage or electrophysiological dysfunction. Patients received 1 point for each component of the CDI for a score range of 0–3. Cardiac dysfunction was defined as a CDI score of ≥1.

Statistical Analysis

We compared differences in the baseline demographic and clinical characteristics of patients with (CDI score ≥1) and without cardiovascular dysfunction (CDI=0) using t-tests and chi-square tests for continuous and categorical variables, respectively. Differences between ordinal variables (ISS, GCS, mGCS) were compared using the Wilcoxon rank-sum test. To assess the relationship between severity of brain injury and cardiac dysfunction, we utilized an ordinal regression model, considering severity of brain injury on an ordinal scale (with mGCS 1 to 6). We used mGCS since landmark studies in TBI have shown total GCS to be unreliable due to the fact that patients who have been intubated, sedated, or intoxicated have unreliable assessments of their verbal or eye-opening responses in the total GCS assessment [18,19]. We adjusted our multivariable regression model for patient’s age, sex, and history of CVD. The Brant test was used to verify the proportional odds assumption. Next, we carried out a survival analysis using the Kaplan-Meier estimate, and compared in-hospital survival rates for those with and without CVD-related complications with the logrank test to examine the significance of differences in estimated survival curves. A multivariable adjusted Cox regression analysis was utilized to examine differences in hospital survival according to CDI score while adjusting for patient’s age, sex, history of CVD, and other known predictors of mortality in patients with TBI, such as pupillary reactivity, Marshall CT-score, pre-hospital hypoxia or hypotension, and mGCS [18]. Adjusted hazard ratios (HRs) with accompanying 95% confidence intervals were calculated. Model assumptions were tested with Schoenfield residuals and goodness-of-fit tests. Data analysis was performed using STATA version 14.0 (Stata Corporation, Bryan, TX).

RESULTS

A total of 502 patients were enrolled in the OPTIMISM study between November 2009 and January 2017. The mean age of the study population was 52 years and approximately 70% of patients were men. The most common cause of TBI in this population was a fall (48%) followed by a motor vehicle accident (19%). We excluded patients who presented with any chest trauma (n= 154), because we were interested in isolating cardiac injury from TBI without the influence of blunt cardiac injury. This resulted in a total of 326 patients who were included in the present analyses (Figure 1).

Figure 1: Study flow diagram.

Patient flow diagram showing the number of patients analyzed for in-hospital survival. CDI = cardiac dysfunction index.

In examining differences in selected baseline demographic and clinical characteristics between those with and without in-hospital cardiac dysfunction (new arrhythmia, PE, DVT, MI, cardiac arrest, EKG or TTE abnormalities), patients with cardiac dysfunction were, on average, 6 years older, were more likely to have a history of CVD, and were more likely to experience pre-hospital hypotension (Table 1). There were 225 patients with EKG data available, 108 patients with TTE data available, and 170 patients with troponin data available.

Table 1:

Patient demographic and clinical characteristics at the time of ICU admission for msTBI.

Baseline Characteristics	Cardiac Dysfunction (n=154)	No Cardiac Dysfunction (n=172)	p-value
Age, mean (SD)	59.4 (21.2)	53.5 (21.2)	0.01
Male	71.4%	68.6%	0.58
Race			0.30
White	88.3%	91.2%
Black	3.9%	5.3%
Asian	2.6%	1.8%
American Indian or Alaskan Native	0.6%	0.6%
Unknown/Not Documented	4.5%	1.2%
Cause of TBI			0.07
Motor Vehicle Accident (MVA)	10.4%	8.7%
Vehicle vs. Pedestrian	1.9%	9.3%
Fall	62.3%	57.6%
Assault	4.5%	8.7%
Gunshot	11.7%	5.8%
Motorcycle/Scooter Accident	3.2%	3.5%
Other	5.7%	6.4%
Total GCS, median (IQR)	5 (5)	7 (3)	0.02
Motor GCS, median (IQR)	3 (4)	4 (2)	0.03
Initial Marshall CT Classification			0.66
Diffuse Injury Type II	54.5%	57.6%
Diffuse Injury Type III	9.7%	11.0%
Diffuse Injury Type IV	7.8%	4.7%
Type 6 (Non-Evacuated Mass Lesion)	26.9%	26.7%
ISS, median (IQR)	26 (9)	22 (13)	0.03
Cardiovascular History			0.03
No Reported History	44.2%	55.8%
Arrhythmia	2.6%	0.6%
CHF	0.6%	0.6%
Hypertension	33.1%	34.9%
Myocardial Infarction	2.6%	1.2%
Coronary Artery Disease	16.9%	7.0%
Pupillary Reactivity			0.49
None	26.6%	22.7%
1 Pupil	9.1%	7.0%
Both Pupils	64.3%	70.3%
Hypotensive in Field/ED	13.6%	4.1%	0.002
Hypoxic in Field/ED	7.8%	4.1%	0.15

ISS = injury severity score. Cardiac dysfunction defined as patients with EKG, sTnI, or TTE derangements as well as those with MI, DVT, PE, new arrhythmia, or cardiac arrest in the ICU.

Frequency of ICU Cardiovascular Complications

The frequency of major cardiovascular complications and abnormal TTE and EKG findings are shown in Table 2. We found that approximately 1 in every 7 patients with msTBI experienced a new cardiac arrhythmia during their neuroICU stay. Approximately two thirds of patients had an EKG abnormality, one third had a TTE abnormality, and half of patients had elevations in their serum troponin levels (Table 2). In terms of missing data, 45% of patients had EKG data available, 22% had information on TTE available, and 35% had sTnI levels available. This left a total of 68 patients with complete data (sTnI, TTE, and EKG) for the calculation of a CDI score (Table 2).

Table 2:

Frequency of cardiac complications and measures of cardiac dysfunction.

Clinical Characteristic/Complication	Value 326
New Arrhythmia During ICU Course	48 (14.7%)
AFib with RVR	11 (3.4%)
Atrial Flutter	7 (2.1%)
Sinus Tachycardia	14 (4.3%)
Symptomatic Bradycardia	11 (3.4%)
Ventricular Fibrillation	2 (0.6%)
Ventricular Tachycardia	8 (2.5%)
EKG Abnormality	93 (62.8%)
Widened QRS	11 (7.7%)
Prolonged QTc Interval	63 (42.9%)
T Wave Inversion	49 (34.0%)
Prolonged PR Interval	5 (3.8%)
Hypotension (requiring vasopressors)	139 (42.6%)
Myocardial Infarction	4 (1.2%)
Pulmonary Embolism	2 (0.6%)
Deep Venous Thrombosis	8 (2.5%)
ICU Cardiac Arrest	25 (7.7%)
TTE Abnormality	23 (32%)
Troponin-I Elevation > 0.05	60 (52.6%)
CDI
0	10 (15%)
1	16 (24%)
2	25 (37%)
3	17 (25%)

Afib RVR = Atrial fibrillation with rapid ventricular response. CDI = cardiac dysfunction index. TTE = transthoracic echocardiography. EKG abnormality = T-wave inversion, R-R interval regularity, QRS length, QTc interval length, axis deviation. TTE abnormality = ejection fraction < 50%, hypokinesis, reduced fractional shortening.

Relationship Between CDI and motor GCS

After adjustment for several potentially confounding demographic (age and sex) and clinical factors (history of CVD), patients with higher CDI scores were significantly more likely to have lower mGCS scores (OR 0.76; 95%CI 0.58–0.99).

In-hospital Survival

Of the 326 patients, 144 died during their acute hospitalization. 35 deaths were due to brain death, 10 deaths were due to cardiac arrest, and 99 deaths were due to withdrawal of life-sustaining treatment. In the unadjusted analysis, patients who developed either EKG, sTnI, or TTE abnormalities, hypotension requiring vasopressors, or had a cardiac arrest in the ICU were shown to have a significantly reduced in-hospital survival in comparison with those who did not experience these complications (Figure 2). On the other hand, patients who developed a new arrhythmia in the ICU experienced similar survival rates compared with those who did not experience a new cardiac arrhythmia (Figure 2).

Figure 2: Kaplan-Meier curve analyses of in-hospital survival.

A KM curve stratified by presence or absence of EKG abnormalities (T-wave inversion, R-R interval regularity, QRS length, QTc interval length, axis deviation). B Echocardiography abnormalities (ejection fraction < 50%, hypokinesis, reduced fractional shortening). C hypotension requiring pharmacologic intervention. D new ICU arrhythmia. E Serum Troponin-I elevation > 0.05. F ICU cardiac arrest.

Univariate Cox regression showed CDI and mGCS to be independently associated with in-hospital survival with hazard ratios of 2.26 (95% CI 1.39–3.70) and 0.73 (95% CI 0.64–0.84), respectively. In the multivariable regression analysis (adjusted for patient’s age, sex, history of CVD, Marshall CT-score, pupillary response, pre-hospital hypoxia and hypotension, and mGCS), ICU survival was significantly lower for those with elevated CDI scores (HR 2.41; 95% CI 1.29–4.53; Figure 2). The effect estimate for mGCS became non-significant when CDI was included in the full regression model with an adjusted HR of 1.14 (95% CI 0.83–1.56).

Since a number of drugs used in the ICU potentially lengthen the QTc interval (e.g., Propofol, proton pump inhibitors), we performed a sensitivity analysis excluding the QTc from the EKG parameters and CDI score. After this exclusion, higher CDI scores continued to be associated with a reduced in-hospital survival (multivariable adjusted HR 2.68; 95%CI 1.54–4.67).

DISCUSSION

The results of this pilot study suggest that cardiac dysfunction is common in hospitalized patients with msTBI and is independently associated with lower in-hospital survival. We also showed an independent association between lower mGCS and greater cardiovascular injury, indicating the potential for a dose-response relationship between degree of brain injury and level of cardiac dysfunction. These findings not only lend further support to a post-msTBI “cardio-cerebral syndrome,” but also show that such a syndrome can be observed with a combination of routine clinical testing (EKG, TTE, sTnI) to potentially prognosticate important clinical outcomes and identify new therapeutic targets for their management.

Frequency of Cardiovascular Complications

We found that approximately two-thirds of the 326 study patients had an EKG abnormality, one third had TTE abnormalities (e.g., low EF, regional or global hypokinesis, fractional shortening), and slightly more than one-half had elevated sTnI levels. These findings are similar to a study in 50 patients with a severe TBI who were hospitalized in a single Egyptian hospital in which 62% of patients had abnormal EKG findings, 54% had elevated troponin I levels, and 28% of patients had TTE abnormalities (e.g., reduced EF, fractional shortening) [12]. These results suggest that cardiac injury is a common sequela of msTBI, affecting more than one half of these critically-ill patients.

Relationship Between CDI and motor GCS

It is unclear in describing brain-heart interactions whether cardiac injury is related to a direct injury to a specific location within the brain (e.g., a “cardunculus”), or if it is simply a marker of severe physiologic stress secondary to the traumatic cerebral insult. Previous studies have examined different measures of cardiac dysfunction, such as systolic dysfunction or the development of new cardiac arrhythmias, in the setting of cerebral insults such as ischemic stroke and subarachnoid hemorrhage (SAH); however, the mechanisms underlying these associations remain poorly understood and are likely multifactorial [20–23]. One small prospective study found an association between GCS scores and degree of systolic dysfunction in 32 patients with msTBI [11].

In the present study, we showed a significant association between the severity of TBI, as measured by mGCS scores, and the degree of cardiovascular dysfunction as measured by the CDI. Individuals with more severe TBI at the time of hospital admission were shown to have an increased likelihood of developing cardiac injury as reflected by higher CDI scores. Furthermore, the hazard ratio effect estimates for mGCS became non-significant when the CDI score was incorporated in the Cox regression model, thereby indicating that CDI may be acting as a mediator. In other words, our data show a significant portion of the total effect of the exposure (TBI) and the outcome (death) potentially operates through a mediator, namely cardiovascular dysfunction. This suggests a potential pathway between the brain and heart in the setting of TBI and give further credence to a “cardio-cerebral syndrome.” This is the first study to suggest that survival in patients with msTBI may be due to effects that are mediated through the cardiovascular system. Future studies should aim to perform causal mediation analyses to confirm these findings.

In-Hospital Survival

A question that remains among patients with the “cardio-cerebral syndrome” is whether cardiac injury is associated with reduced survival in patients with msTBI. In other words, is the brain injury so severe that any observed cardiac injury is of little importance to patient’s short-term survival? Our data showed a significant reduction in survival during the acute hospitalization among patients with increased makers of cardiac dysfunction and higher levels of cardiac injury compared to those without evidence of reduced cardiac function, even after adjusting for several variables of prognostic importance. This is similar to the results of a previous small study which found derangements in troponin, echocardiography, and systolic blood pressure to be associated with increased mortality in a cohort of 50 patients with severe TBI (GCS <8) [12]. There are also limited data to suggest that protecting the heart with beta blockers in patients with TBI may improve patients’ short-term survival, presumably due to their cardio-protective effects against the adrenergic surge commonly noted after a severe brain injury [24,25]. Future research should validate our findings, which might eventually prove useful in finding novel targets for therapeutic intervention in patients with TBI.

Study Strengths and Limitations

To the best of our knowledge, this is the first study to highlight the potential for cardiac injury as a mediator in the causal pathway between traumatic brain injury and death. We used a rigorously and prospectively collected longitudinal patient cohort from the OPTIMISM study, in which cardiac complications are collected pre-hoc. Our CDI score provides a simple and clinically meaningful parameter to objectively measure cardiac dysfunction in this patient population. For the purposes of this exploratory analysis, we decided to equally weight each of the CDI components, which provided a more simplified stratification for patients with varying levels of cardiac injury (score 0–4) and would allow us to more directly compare patients across different strata of cardiac injury. We recognize the need for future prospective studies to consider prognostication models with different weighting assigned to each of the individual components included in any such model.

Our study has the limitations of an observational study design with retrospective data collection. ECG, TTE and sTnI findings were not performed routinely on all patients, since these tests were ordered at the discretion of the clinical team based on clinical necessity or symptoms limiting the CDI analysis cohort to a small number. In comparing groups with and without cardiac dysfunction (Table 1), it is important to note that these comparisons are somewhat limited as the “no dysfunction” group may have in fact had some cardiac dysfunction present but did not have either and EKG, sTnI, or TTE performed during their hospitalization. We did not collect EKG information prior to the traumatic event, so it is difficult to determine what proportion of the EKG changes were new. Future studies should prospectively and serially examine changes in these clinical variables over the course of the patient’s hospitalization to provide more robust and interpretable data. The CDI in our study was applied to a relatively small number of subjects (68) who had all of the available cardiac testing. This group likely represents a higher-risk subset of patients with baseline CAD or other cardiac risk factors. The CDI could prove to be helpful in prognosticating outcomes for this group, but future research with a more systematic collection of cardiac data could help broaden the generalizability of the CDI to include younger patients without extensive cardiovascular risk factors.

Our study provides preliminary, hypothesis generating data which should be validated by a larger prospective study with systematic cardiac data collection. Furthermore, our study lacked the examination of long-term outcome data, which could be part of future research to examine the effects of cardiac injury on long-term outcomes in TBI patients.

CONCLUSIONS

Our study lays a foundation for a novel, potentially useful tool in prognosticating outcomes in patients with msTBI. This is important because we are currently hindered by a paucity of treatment modalities in caring for these critically-ill patients and a limited knowledge base in terms of which non-neurological clinical parameters affect patient outcomes. The results of our pilot study, if validated, suggest an independent association between the severity of brain injury and development of cardiac injury in patients hospitalized in a neuroICU with msTBI. The level of cardiac dysfunction, which we have shown can be easily measured with the CDI scoring tool, was independently associated with reduced in-hospital survival. Larger studies are needed to prospectively and more systematically examine various cardiac findings among patients hospitalized with msTBI. Further research is also needed to corroborate our findings and validate whether adverse outcomes after msTBI are mediated by cardiac injury, which, in turn, could pose a target for novel prognostication tools and therapeutic interventions.

Figure 3: Survival function of Cox proportional hazards regression across 4 levels of cardiac injury.

Cardiac dysfunction index = 0 indicates there were no EKG abnormalities (T-wave inversion, R-R interval regularity, QRS length, QTc interval length, axis deviation), Echocardiography abnormalities (ejection fraction < 50%, hypokinesis, reduced fractional shortening), or serum Troponin-I elevations (>0.05ng/mL). Cardiac dysfunction index of 3 indicates all three abnormalities present.

Highlights:

• Cardiac injury is common among patients with traumatic brain injury (TBI)

• Patients with more severe brain injuries have worse cardiac injury

• Cardiac dysfunction is associated with a reduction in in-hospital survival

• Death from TBI is potentially mediated by the cardiovascular system